Confused over Conda Package manager vs PIP

[JanOlvegg]

Both Conda and PIP appears to be package manager. From what I can read so far, PIP Is specific to Python but Conda is more language agnostic. Is there a reason to favor one package manager over another?

[snippsat]

Conda can in cases when eg need a compiler be much easier to use as it will download what's needed.
Conda can also install Python versions.pip can not do that
As i showed in your other Thread so will conda install -c conda-forge hnswlib fix the compiler issues that pip can not do.

Basic stuff with Miniconda,as you see so dos it come with both pip and conda.

# Activate
C:\code\miniconda3\Scripts
λ activate.bat

C:\code\miniconda3\Scripts
(base) λ cd..

C:\code\miniconda3
(base) λ python -V
Python 3.10.9

C:\code\miniconda3\Scripts
(base) λ pip -V
pip 22.3.1 from C:\code\miniconda3\lib\site-packages\pip (python 3.10)

C:\code\miniconda3\Scripts
(base) λ conda -V
conda 23.1.0

Always activate a environment,and i would advice not using (base) environment,make a new one.
Then if trouble can just delete that environment and try again.

# Make a environment this will also install Python 3.10.9 and pip
conda create --name my_env -c conda-forge spyder jupyterlab pandas python=3.10.9

When finish activate (my_env) the new environment and install other stuff that needed.

[jefsummers]

And, if you prefer a more graphical interface for your environment management, conda offers Anaconda. Anaconda allows you to create and manage environments and integrates well with most development packages (VS Code, Jupyter, Spyder, PyCharm, etc.)

[JanOlvegg]

Thanks,

So I played around with Anaconda. The main difference between Conda and PIP appears to be the scope. Pip controls what packages for the python version you install. You can use pip and venv to control what packages go a project, but you can't use pip to install different version of pythion. Third party code/wheels appear to come uncompile.

Conda exists a level above python and allow you to setup different python release locally. Conda itself seems run locally, too. Packages are all pre-compile. There seems to be two variant Anaconda which includes a boatload of package so you don't have to worry about including packages and miniconda which has a minimal set of packages.

One issue so far with Conda is that the software is not free to corporations (of over 200). If I want to install it at work, I will have to do a lot of paperwork to expense a purchase. This can be both good and bad. Paid software means support which would check some support requirement boxes, but getting it install in the first place would be a super pain.